Fix for gnus-registry on newer Emacsen

Fix for gnus-registry on newer Emacsen

[David Engster]

Probably no one noticed, but if you use the registry on the current
pretest or trunk, you'll get a message on startup about an "Unsafe call
to eieio-persistent-read".

I'll just pushed a fix for this, but of course we have to stay
compatible with older Emacsen, so I had to wrap this in a condition-case
and an additional with-no-warnings. However, it seems the latter doesn't
have any effect on XEmacs, so the buildbot is red there. Does anyone
know what can be done here?

-David

Re: Fix for gnus-registry on newer Emacsen

[Dave Abrahams]

on Fri Feb 22 2013, David Engster <deng-AT-randomsample.de> wrote:

> Probably no one noticed, but if you use the registry on the current
> pretest or trunk, you'll get a message on startup about an "Unsafe call
> to eieio-persistent-read".

Oh, I noticed!

> I'll just pushed a fix for this, 

Thanks!

> but of course we have to stay compatible with older Emacsen, so I had
> to wrap this in a condition-case and an additional
> with-no-warnings. However, it seems the latter doesn't have any effect
> on XEmacs, so the buildbot is red there. Does anyone know what can be
> done here?

No clue, sorry.

-- 
Dave Abrahams

Re: Fix for gnus-registry on newer Emacsen

[David Engster]

David Engster writes:
> I'll just pushed a fix for this, but of course we have to stay
> compatible with older Emacsen, so I had to wrap this in a condition-case
> and an additional with-no-warnings. However, it seems the latter doesn't
> have any effect on XEmacs, so the buildbot is red there. Does anyone
> know what can be done here?

I've now silenced the XEmacs byte compiler for argument mismatches for
the whole file. The persistent registry does not work on XEmacs anyway,
so IMO this is not a problem, but if someone thinks otherwise and has a
better solution, I'm all ears.

-David

Re: Fix for gnus-registry on newer Emacsen

[Uwe Brauer]

[-- Attachment #1: Type: text/plain, Size: 969 bytes --]

>> "David" == David Engster <deng@randomsample.de> writes:

   > David Engster writes:

   >> I'll just pushed a fix for this, but of course we have to stay
   >> compatible with older Emacsen, so I had to wrap this in a
   >> condition-case and an additional with-no-warnings. However, it
   >> seems the latter doesn't have any effect on XEmacs, so the
   >> buildbot is red there. Does anyone know what can be done here?

   > I've now silenced the XEmacs byte compiler for argument mismatches
   > for the whole file. The persistent registry does not work on XEmacs
   > anyway, so IMO this is not a problem, but if someone thinks
   > otherwise and has a better solution, I'm all ears.

Oops, using Xemacs.21.5X, registered (persistent) registry worked for me
in earlier version of gnus, say between 1 and 2 years old, but for
months I tried to get them to work the way it used to be. Are you
telling me that this feature does not work for new versions in Xemacs?

[-- Attachment #2: smime.p7s --]
[-- Type: application/pkcs7-signature, Size: 5556 bytes --]

Re: Fix for gnus-registry on newer Emacsen

[David Engster]

Uwe Brauer writes:
>>> "David" == David Engster <deng@randomsample.de> writes:
>    > I've now silenced the XEmacs byte compiler for argument mismatches
>    > for the whole file. The persistent registry does not work on XEmacs
>    > anyway, so IMO this is not a problem, but if someone thinks
>    > otherwise and has a better solution, I'm all ears.
>
> Oops, using Xemacs.21.5X, registered (persistent) registry worked for me
> in earlier version of gnus, say between 1 and 2 years old, but for
> months I tried to get them to work the way it used to be. Are you
> telling me that this feature does not work for new versions in Xemacs?

Yes. Emacs supports printing and reading hash-tables since version 23.2
or so, and Ted switched to that when he rewrote the registry back end.

-David